1748259 Members
3625 Online
108760 Solutions
New Discussion юеВ

fbackup

 
nibble
Super Advisor

fbackup

Guys, our customer has a running MSL6000 with an Ultrium 2-SCSI tape drive. The script that they used in backup includes everything "/" and excludes only /cdrom which I found not really appropriate, but it was working for a long time. recently, their backup failed with the ff observation:
a. successful backup was able to write 543177+ files, basing on the wc -l index file
b. when failed, it was able to write up to 290000 (+ - 10%) files and it seems to freez.
c. successful backup 1-1.5 hrs only
d. unsuccssful freezes, and when killed the number of files written is stated on b.
e. error when killed (which is normal i believe)
292929: /data2/test/rdg/RDISTOVER0.dat 2
292930: /data2/test/rdg/RDISTOVER0.idx 6
292931: /data2/test/rdg/REP5BYP6.dat 2207538
fbackup(1027): Backup did not complete : Reader or Writer process exit
fbackup(3055): total file blocks read for backup: 2745794013
fbackup(3056): total blocks written to output file /dev/rmt/1m: 3842111423
f. config file is set as follows:
blocksperrecord 128
checkpointfreq 256
this is working for a long time..i think theres no issue
g. I just realized that they mentioned they replaced the power supply of the server connected to this, I think they might have disturbed the cable connection of the scsi, etc..
h. tar works fine with smaller number of files, however I have to test if tar works with at least 300000 files
i. no significant increase of the file sizes
j. another error that might help:
fbackup(1004): session begins on Tue Nov 18 22:00:00 2008
fbackup(1517): /net not backed up - 'n' option (NFS) not specified
fbackup(3203): volume 1 has been used 11 time(s)
fbackup(3024): writing volume 1 to the output file /dev/rmt/1m
fbackup(3013): WRITE ERROR while writing data record, at media record 16776865
fbackup(3102): attempting to make this volume salvagable
fbackup(3123): could not read the previous checkpoint record
fbackup(3105): writing 2 EOFs and rewinding the tape
fbackup(3106): please mount a good tape
fbackup(3310): enter '^[yY]' when volume 1 is ready on /dev/rmt/1m,
or '^[nN]' to discontinue:
fbackup(3004): writer aborting
fbackup(1002): Backup did not complete : Reader or Writer process exit
(ive search this one, and more on the drive or cable)
k. no scsi errors in the OS, no errors in drive based on MSL6000 remote console, tape media has been replaced several times with new ones

Please see what you think if im on the right track
1. There shoule be no limitation with the size of the backup whether 500,000 files or up..and it worked before
2. Might be a good idea to re-seat the cables, they might had did something when replacing the psu of the server
3. any other test to do??

OS: HPUX 11.11
HW: RP4440
MSL6000
4 REPLIES 4
TTr
Honored Contributor

Re: fbackup

Checking the cables is definitely a good idea. Do you know if they disconnected the SCSI cable(s) or only pushed them out of the way?

Another thing you should try is to use a brand new tape (or two) to see if this is a tape related problem.

And lastly you should check if you have problems with the tape drive. It may be failing after it has run for a while from overheating etc.
nibble
Super Advisor

Re: fbackup

Thanks TTR. the tapes are brand new. I dont have any physical access to the server, but based on the OS logs and MSL6000 console, there's no error on the drive itself. As for the cables, thats what Im thinking too. I cant do an onsite now.
OldSchool
Honored Contributor

Re: fbackup

what ttr said. if you haven't tried a new tape (yet) make that option one, as the message about write error / two end of files leads me to believe it might be the culprit.

after the backups are working, you might want to look at your fbackup config settings.

see the post here about improving performance:
http://forums13.itrc.hp.com/service/forums/questionanswer.do?admit=109447627+1227214316703+28353475&threadId=733227
nibble
Super Advisor

Re: fbackup

Yes guys, I'll try your suggestions.